Refactor, cleanup and minor fixes

.github/workflows/conda-build.yml

@@ -19,16 +19,15 @@ jobs:
     - uses: actions/checkout@v4
     - uses: conda-incubator/setup-miniconda@v3
       with:
+        miniforge-version: latest
         auto-update-conda: true
         auto-activate-base: true
         activate-environment: ""
-        channel-priority: strict
         conda-remove-defaults: "true"
-        miniforge-variant: Miniforge3
     - shell: bash -l {0}
       run: conda info --envs
     - name: Build pytorch-3dunet
       shell: bash -l {0}
       run: |
-        conda install -q conda-build
+        conda install -q conda-build conda=25.7.0
         conda build -c conda-forge conda-recipe

.github/workflows/lint.yml

@@ -0,0 +1,23 @@
+name: Lint
+
+on:
+  push:
+    branches:
+      - "master"
+  pull_request:
+    branches: [master]
+
+jobs:
+  ruff:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+      - name: Install ruff
+        run: pip install ruff
+      - name: Run ruff check
+        run: ruff check .
+      - name: Run ruff format check
+        run: ruff format --check .

README.md

@@ -37,43 +37,48 @@ The format of the raw and label datasets depends on whether the problem is 2D or
 | multi-channel  | (C, 1, Y, X) | (C, Z, Y, X) |
 
 ## Prerequisites
-
-- NVIDIA GPU
-- CUDA CuDNN
+- [Miniconda](https://www.anaconda.com/docs/getting-started/miniconda/install)
+- Python 3.11+
+- NVIDIA GPU (optional but recommended for training/prediction speedup)
 
 ### Running on Windows/OSX
 
 `pytorch-3dunet` is a cross-platform package and runs on Windows and OS X as well.
 
 ## Installation
 
-- The easiest way to install `pytorch-3dunet` package is via conda:
+The easiest way to install `pytorch-3dunet` package is via conda:
 
 ```bash
-conda install -c conda-forge pytorch-3dunet
-```
+# Created new conda environment "3dunet" with the latest python version from the conda-forge channel
+conda create -n 3dunet python -c conda-forge -y
 
-**Note:** The conda package does not include PyTorch dependencies. You need to install them separately in your conda environemt:
+# Activate the conda environment
+conda activate 3dunet
 
-```bash
-# Install PyTorch with CUDA support (adjust for your CUDA version, below it's CUDA 11.8)
-pip install torch torchvision --index-url https://download.pytorch.org/whl/cu118
+# pytorch-3dunet does not include PyTorch dependencies, so that one can install the desired PyTorch version (with/without CUDA support) separately
+pip install torch torchvision
+# you may need to adjust the command above depending on your GPU and the CUDA version you want to use, e.g. for CUDA 12.6:
+# pip install torch torchvision --index-url https://download.pytorch.org/whl/cu126
+# or for CPU-only version:
+# pip install torch torchvision --index-url https://download.pytorch.org/whl/cpu
 
-# Or install CPU-only version
-pip install torch torchvision --index-url https://download.pytorch.org/whl/cpu
+# Install the latest pytorch-3dunet package from conda-forge channel
+conda install -c conda-forge pytorch-3dunet
 ```
 
 After installation the following commands will be accessible within the conda environment:
 `train3dunet` for training the network and `predict3dunet` for prediction (see below).
 
-- One can also install directly from source, i.e. go to the checkout directory and run:
-
+One can also install directly from source, i.e. go to the checkout directory and run:
 ```
 pip install -e .
 ```
 
 ### Installation tips
-Make sure that the installed `torch` is compatible with your CUDA version, otherwise the training/prediction will fail to run on GPU. 
+PyTorch package comes with their own CUDA runtime libraries, so you don't need to install CUDA separately on your system.
+However, you must ensure that the PyTorch/CUDA version you choose is compatible with your GPU’s compute capability. 
+See [PyTorch installation guide](https://pytorch.org/get-started/locally/) for more details. 
 
 ## Train
 Given that `pytorch-3dunet` package was installed via conda as described above, you can train the network by simply invoking:
@@ -98,7 +103,7 @@ One can monitor the training progress with Tensorboard `tensorboard --logdir <ch
    When training with `WeightedCrossEntropyLoss`, `CrossEntropyLoss` the target dataset has to be 3D, see also pytorch
    documentation for CE loss: https://pytorch.org/docs/master/generated/torch.nn.CrossEntropyLoss.html
 2. When training with `BCEWithLogitsLoss`, `DiceLoss`, `BCEDiceLoss`, `GeneralizedDiceLoss` set `final_sigmoid=True` in
-   the `model` part of the config so that the sigmoid is applied to the logits.
+   the `model` part of the config so that the sigmoid is applied to the logits in inference mode.
 3. When training with cross entropy based losses (`WeightedCrossEntropyLoss`, `CrossEntropyLoss`) set
    `final_sigmoid=False` so that `Softmax` normalization is applied to the logits.
 
@@ -303,6 +308,7 @@ pip install -e .
 
 Tests can be run via `pytest`.
 The device the tests should be run on can be specified with the `--device` argument (`cpu`, `mps`, or `cuda` - default: `cpu`).
+Linting is done via `ruff` (see `pyproject.toml` for configuration).
 
 ## Release new version on `conda-forge` channel
 To release a new version of `pytorch-3dunet` on the `conda-forge` channel, follow these steps:

environment.yaml

@@ -15,6 +15,7 @@ dependencies:
   - scikit-image
   - pyyaml
   - pytest
+  - ruff
   - pip
   - pip:
     - torch

pyproject.toml

@@ -37,4 +37,26 @@ predict3dunet = "pytorch3dunet.predict:main"
 version = {attr = "pytorch3dunet.__version__.__version__"}
 
 [tool.setuptools.packages.find]
-exclude = ["tests*"]
+exclude = ["tests*"]
+
+[tool.ruff]
+line-length = 120
+target-version = "py311"
+
+[tool.ruff.lint]
+select = [
+    "E",   # pycodestyle errors
+    "W",   # pycodestyle warnings
+    "F",   # pyflakes
+    "I",   # isort
+    "B",   # flake8-bugbear
+    "C4",  # flake8-comprehensions
+    "UP",  # pyupgrade
+]
+ignore = [
+    "E501",  # line too long (handled by formatter)
+]
+
+[tool.ruff.format]
+quote-style = "double"
+indent-style = "space"

pytorch3dunet/__init__.py

@@ -1 +1 @@
-from .__version__ import __version__
+from .__version__ import __version__  # noqa: F401

pytorch3dunet/__version__.py

@@ -1 +1 @@
-__version__ = '1.9.3'
+__version__ = "1.9.3"

pytorch3dunet/augment/transformer-test.ipynb

@@ -6,15 +6,26 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "from transforms import LabelToAffinities, RandomFlip, RandomRotate, RandomRotate90, RandomContrast, Normalize, ElasticDeformation, StandardLabelToBoundary, AdditiveGaussianNoise, AdditivePoissonNoise, LabelToBoundaryAndAffinities, FlyWingBoundary, LabelToBoundaryAndAffinities, Standardize\n",
-    "import matplotlib.pyplot as plt\n",
     "import h5py\n",
+    "import matplotlib.pyplot as plt\n",
     "import numpy as np\n",
+    "from transforms import (\n",
+    "    AdditiveGaussianNoise,\n",
+    "    AdditivePoissonNoise,\n",
+    "    ElasticDeformation,\n",
+    "    LabelToAffinities,\n",
+    "    RandomContrast,\n",
+    "    RandomFlip,\n",
+    "    RandomRotate,\n",
+    "    RandomRotate90,\n",
+    "    Standardize,\n",
+    "    StandardLabelToBoundary,\n",
+    ")\n",
     "\n",
-    "path = '../../resources/sample_ovule.h5'\n",
-    "with h5py.File(path, 'r') as f:\n",
-    "    raw = f['raw'][...]\n",
-    "    label = f['label'][...]"
+    "path = \"../../resources/sample_ovule.h5\"\n",
+    "with h5py.File(path, \"r\") as f:\n",
+    "    raw = f[\"raw\"][...]\n",
+    "    label = f[\"label\"][...]"
    ]
   },
   {
@@ -168,28 +179,28 @@
    ],
    "source": [
     "# LabelToBoundary demo\n",
-    "path = '../resources/sample_patch.h5'\n",
+    "path = \"../resources/sample_patch.h5\"\n",
     "\n",
-    "with h5py.File(path, 'r') as f:\n",
-    "    label = f['label'][...]\n",
-    "    t1 = LabelToAffinities(offsets=[1,4,6,8], aggregate_affinities=True)\n",
+    "with h5py.File(path, \"r\") as f:\n",
+    "    label = f[\"label\"][...]\n",
+    "    t1 = LabelToAffinities(offsets=[1, 4, 6, 8], aggregate_affinities=True)\n",
     "    label_transformed1 = t1(label)\n",
-    "    \n",
+    "\n",
     "    sltb = StandardLabelToBoundary(blur=True, sigma=1.3)\n",
     "    label_transformed1 = np.append(label_transformed1, sltb(label), axis=0)\n",
-    "    \n",
-    "    fig, axes = plt.subplots(1, 6, figsize=(25,  5))\n",
+    "\n",
+    "    fig, axes = plt.subplots(1, 6, figsize=(25, 5))\n",
     "    ax = axes.ravel()\n",
     "\n",
     "    # show label\n",
-    "    ax[0].set_title('Label')\n",
-    "    ax[0].imshow(label[40, ...], cmap='flag')\n",
+    "    ax[0].set_title(\"Label\")\n",
+    "    ax[0].imshow(label[40, ...], cmap=\"flag\")\n",
     "    ax[0].set_axis_off()\n",
     "\n",
-    "    for i in range(label_transformed1.shape[0]):        \n",
-    "        ax[i+1].set_title(f'Affinities{i}')\n",
-    "        ax[i+1].imshow(label_transformed1[i, 40, ...])\n",
-    "        ax[i+1].set_axis_off()"
+    "    for i in range(label_transformed1.shape[0]):\n",
+    "        ax[i + 1].set_title(f\"Affinities{i}\")\n",
+    "        ax[i + 1].imshow(label_transformed1[i, 40, ...])\n",
+    "        ax[i + 1].set_axis_off()"
    ]
   },
   {
@@ -215,36 +226,44 @@
    "source": [
     "# demo some of the augmentations\n",
     "rs = np.random.RandomState()\n",
-    "raw_transformers = [RandomFlip(rs), RandomRotate90(rs), RandomRotate(rs, angle_spectrum=20, axes=[(2, 1)], mode='reflect'), RandomContrast(rs, alpha=(0.5, 1.5), execution_probability=1.0), ElasticDeformation(rs, 3, alpha=20, sigma=3, execution_probability=1.0), AdditiveGaussianNoise(rs, scale=(0.0, 0.5), execution_probability=1.0), AdditivePoissonNoise(rs, lam=(0.0, 0.5), execution_probability=1.0)]\n",
+    "raw_transformers = [\n",
+    "    RandomFlip(rs),\n",
+    "    RandomRotate90(rs),\n",
+    "    RandomRotate(rs, angle_spectrum=20, axes=[(2, 1)], mode=\"reflect\"),\n",
+    "    RandomContrast(rs, alpha=(0.5, 1.5), execution_probability=1.0),\n",
+    "    ElasticDeformation(rs, 3, alpha=20, sigma=3, execution_probability=1.0),\n",
+    "    AdditiveGaussianNoise(rs, scale=(0.0, 0.5), execution_probability=1.0),\n",
+    "    AdditivePoissonNoise(rs, lam=(0.0, 0.5), execution_probability=1.0),\n",
+    "]\n",
     "\n",
-    "with h5py.File(path, 'r') as f:\n",
-    "    raw = f['raw'][...]\n",
+    "with h5py.File(path, \"r\") as f:\n",
+    "    raw = f[\"raw\"][...]\n",
     "    mid_z = raw.shape[0] // 2\n",
     "    raw = Standardize(np.mean(raw), np.std(raw))(raw)\n",
-    "    #raw = Normalize(np.min(raw), np.max(raw))(raw)\n",
-    "    label = f['label'][...]\n",
-    "    \n",
+    "    # raw = Normalize(np.min(raw), np.max(raw))(raw)\n",
+    "    label = f[\"label\"][...]\n",
+    "\n",
     "    # show transforms\n",
     "    fig, axes = plt.subplots(len(raw_transformers), 3, figsize=(15, len(raw_transformers) * 5))\n",
     "    ax = axes.ravel()\n",
-    "    \n",
+    "\n",
     "    for t, i in enumerate(range(0, 3 * len(raw_transformers), 3)):\n",
     "        transformer = raw_transformers[t]\n",
-    "        ax[i].set_title('Raw')\n",
+    "        ax[i].set_title(\"Raw\")\n",
     "        ax[i].imshow(raw[mid_z, ...])\n",
     "        ax[i].set_axis_off()\n",
     "\n",
     "        # show boundary for the 1st offset\n",
-    "        ax[i+1].set_title(f'{type(transformer).__name__}-0')\n",
+    "        ax[i + 1].set_title(f\"{type(transformer).__name__}-0\")\n",
     "        aug = transformer(raw)\n",
-    "        ax[i+1].imshow(aug[mid_z, ...])\n",
-    "        ax[i+1].set_axis_off()\n",
+    "        ax[i + 1].imshow(aug[mid_z, ...])\n",
+    "        ax[i + 1].set_axis_off()\n",
     "\n",
     "        # show boundary for the 4th offset\n",
-    "        ax[i+2].set_title(f'{type(transformer).__name__}-1')\n",
+    "        ax[i + 2].set_title(f\"{type(transformer).__name__}-1\")\n",
     "        aug = transformer(raw)\n",
-    "        ax[i+2].imshow(aug[mid_z, ...])\n",
-    "        ax[i+2].set_axis_off()"
+    "        ax[i + 2].imshow(aug[mid_z, ...])\n",
+    "        ax[i + 2].set_axis_off()"
    ]
   },
   {
@@ -273,11 +292,11 @@
     "rs = np.random.RandomState()\n",
     "transformer = AdditiveGaussianNoise(rs, scale=(0.0, 1.0), execution_probability=1.0)\n",
     "\n",
-    "ax[0].set_title('Raw')\n",
+    "ax[0].set_title(\"Raw\")\n",
     "ax[0].imshow(raw[mid_z, ...])\n",
     "ax[0].set_axis_off()\n",
     "for i in range(1, 10):\n",
-    "    ax[i].set_title(f'Noise-{i}')\n",
+    "    ax[i].set_title(f\"Noise-{i}\")\n",
     "    aug = transformer(raw)\n",
     "    ax[i].imshow(aug[mid_z, ...])\n",
     "    ax[i].set_axis_off()"
@@ -309,11 +328,11 @@
     "rs = np.random.RandomState()\n",
     "transformer = AdditivePoissonNoise(rs, lam=(0.0, 1.0), execution_probability=1.0)\n",
     "\n",
-    "ax[0].set_title('Raw')\n",
+    "ax[0].set_title(\"Raw\")\n",
     "ax[0].imshow(raw[mid_z, ...])\n",
     "ax[0].set_axis_off()\n",
     "for i in range(1, 10):\n",
-    "    ax[i].set_title(f'Noise-{i}')\n",
+    "    ax[i].set_title(f\"Noise-{i}\")\n",
     "    aug = transformer(raw)\n",
     "    ax[i].imshow(aug[mid_z, ...])\n",
     "    ax[i].set_axis_off()"
@@ -346,11 +365,11 @@
     "t1 = AdditiveGaussianNoise(rs, scale=(0.0, 1.0), execution_probability=1.0)\n",
     "t2 = AdditivePoissonNoise(rs, lam=(0.0, 0.5), execution_probability=1.0)\n",
     "\n",
-    "ax[0].set_title('Raw')\n",
+    "ax[0].set_title(\"Raw\")\n",
     "ax[0].imshow(raw[mid_z, ...])\n",
     "ax[0].set_axis_off()\n",
     "for i in range(1, 10):\n",
-    "    ax[i].set_title(f'Noise-{i}')\n",
+    "    ax[i].set_title(f\"Noise-{i}\")\n",
     "    aug = t1(t2(raw))\n",
     "    ax[i].imshow(aug[mid_z, ...])\n",
     "    ax[i].set_axis_off()"